<p>I think that the cross-admit studies of preferences are not reliable, because a student must have applied to both X and Y in order have any possibility to contribute to the apparent preference for X or Y, via the cross-admit data.</p>
<p>Suppose for example that school H is hard to get into, and that it attracts a certain type of person. A person who prefers school H over any other school will almost certainly apply there, if the person has qualifications in the general range for admission. The person is likely to apply to a few other “top-level” schools as well (A, B, C . . . ). If the person is admitted by H, that person will probably pick H, which thus “wins” over A, B, C . . . </p>
<p>Then suppose that another student prefers A over H, and has H-level qualifications. Will that student apply to H? Perhaps on the East Coast, the answer is “yes,” but in my area of the country, there is a reasonable chance that the student will not apply to H. So the student’s preference for A over H is at least as strong as the H-preference of the hypothetical person in the previous paragraph (and probably stronger), but this person will never show up in a study of preferences by cross-admits.</p>
<p>I think that the reason that Caltech is not doing as well in the cross-admit standings (relying on a poster above for the validity of that) is this: At one time, it was the case that many Caltech applicants really, really wanted to go to Caltech. Recently, Caltech has gained a wider reputation as a school that will admit students with truly outstanding academic qualifications, with reasonable reliability. This contrasts with the admissions practices at many other “top” schools (for good or ill, depending on your point of view). So a student with outstanding academic qualifications who would consider Caltech, but would actually prefer H (or A, B, C . . . ) is probably more likely to apply to Caltech now than he/she would have been in the past.</p>
<p>With regard to the rigor of UC Berkeley vs. MIT, I think that it is essential to consider the curricular choices made by the students–and not simply the choice of major, but the specific course choices. In the natural sciences, there tends to be a very wide range of possible choices of courses in the undergrad program, as well as the option of taking grad courses. The level of rigor that a student encounters depends on the level of the other students in the classes that the student has chosen, and on the expectations of the professors teaching those classes. My university offers about 6,000 courses. There are some very easy routes to a degree, even within the physical sciences. There are also some very challenging routes to a degree. There might be less curricular flexibility in engineering, but I would guess that even there, some options are available that would make the degree more or less challenging. The minimal bar at MIT is probably higher than the minimal bar for the same curriculum at Berkeley. That doesn’t say much about the choices of a really strong student, though.</p>
<p>So what QuantMech says is at least potentially a problem but (i) it seems unlikely to dramatically change the conclusion in the case of MIT v. Berkeley where Berkeley currently gets <5% of cross-admits although it may make things somewhat less lopsided. (ii) In the Avery et al paper they asked students to list preferences for the colleges they were considering and constructed a ranking based on these preferences. The results didn’t change that much. I think this increases our confidence that the revealed preference analysis is not extremely sensitive to these factors</p>
<p>@UMTYMP student, I think that the utility of the cross-admit data between any two universities depends a lot on which two universities are being compared, and the factors that go into the choices.</p>
<p>For some pairs of universities, I think that the “revealed preference” analysis is dominated by other factors than rigor, or quality, or prestige. For others, the “revealed preference” analysis is probably valid as an indicator of rigor, or quality, or prestige.</p>
<p>I recall the surprise of the Harvard authors of the original cross-admit study when they discovered that Caltech was #2 in cross-admit preferences–an outcome that was totally inexplicable to them. It was no surprise to me that Harvard “won.” I think that the nature of the cross-admit study itself, coupled with the nature of Harvard and its applicants, pretty much guaranteed that. Oh, yeah, did I mention that the authors were affiliated with Harvard?</p>
<p>I wasn’t commenting specifically on the cross-admit data for UC Berkeley vs. MIT. However, I am not at all surprised to learn that MIT swamps Berkeley in the cross-admit study, as you have mentioned. MIT has a lot of prestige among the general population, while Berkeley has less among the general population. That has got to sway a lot of the cross-admits. Financial aid at the two schools may factor into the choices. </p>
<p>My Berkeley/MIT comparison comments really address the relative rigor of the two schools. Some routes through Berkeley are more rigorous than some routes through MIT, and some are less. You can’t really tell about the rigor of any given student’s coursework without looking at the transcript, to see the choice of courses. </p>
<p>It does not change the scenario with regard to the H-preference student and the A-preference student, in my post above, if one adds comparisons of colleges that the applicants were “considering.” The H-preference student is still considering A, while the A-preference student is not considering H. So the A-preference student still doesn’t show up in cross-preference analysis. In other regions of the country, there may not be many student who act like the A-preference student I mentioned, but there are plenty around here.</p>
I actually am quite surprised – in the MIT data (at least for the year I did the data analysis), UMich and Georgia Tech get many more (~4x) MIT admits than Berkeley. Given that a greater percentage of the admit pool is from California than either Georgia or Michigan, I would expect Berkeley to peel off more students than it does, particularly among in-staters. Berkeley (25k undergrads) is about the same size as UMich (28k undergrads), but bigger than GT (15k undergrads).</p>
<p>Perhaps GT and UMich offer better financial aid for out-of-state students than UCB is offering? In terms of public universities with great engineering programs, my estimate would be that Berkeley has the “name” edge, which is why I find it puzzling.</p>
<p>I think GT and UMich might be cheaper for in-state students than Berkeley is for in-state for students. It seems plausible that Berkeley at 20k/year cheaper may get many fewer students than GT or UMich at 30-50k/year cheaper. Granted these would be different pools of students as well. There is also a fairly small number of students (like 12 v 3?) so noise may play a part too.</p>
<p>I will respond to the other posts when I get more time.</p>
<p>Fear of earthquakes? (which would not influence the UMich nor GT cross-admit results)
Incidentally, since you are here, molliebatmit, congratulations on completing your doctorate, and also on the NIH grant!</p>
<p>Thanks, QM. The grant’s from some time ago, though – that post about it was very old! I’m now applying (and applying, and applying) for grants and postdoc fellowships in my new lab.</p>
<p>Good lucky with the new grant application, molliebatmit!</p>
<p>On a more serious note than the earthquakes comment, both UMich and GT have a very high level of prestige in their respective states. UMich, for example, seems to me to be more highly regarded within Michigan than Ohio State is in Ohio, or UW Madison in Wisconsin, or UIUC in Illinois–based on my observations in those states, anyway. Similarly for GT, from what I’ve heard. I don’t know whether Berkeley enjoys the edge with the in-state high-school group that UMich and GT do.</p>
<p>Another factor that is somewhat different between the cases: UMich has an exceptionally strong engineering school, which outranks many of its other programs–e.g., the programs in the physical sciences are strong, but they are not nationally ranked as highly as the engineering programs. (I exclude the law school and medical school from this consideration.) GT is engineering focused. UC Berkeley has a strong College of Engineering, but I think that it has very comparably ranked programs in other areas. This might influence whether people think about any particular school as a “good school for engineers.”</p>
<p>Then there are regional characteristics. Are “top” students in California somewhat more likely to go to the East Coast for college than are “top” students in Michigan or Georgia? I don’t know, but this seems at least possible to me. </p>
<p>Personally, Berkeley couldn’t make a strong case to win over MIT in the cross-admit battles. There just isn’t any compelling reason to sway cross-admits to go to Berkeley over MIT. That is also quite true for Berkeley vs Stanford and Harvard. But, I would venture that Berkeley may have a compelling case to win in a cross-admit battle between it and Yale, and possibly, Princeton. Having said that, it’s not conclusive to claim that Berkeley COE peeps are academically “much weaker” peeps compared to MIT peeps. Maybe weaker. But not MUCH weaker. Not all MIT rejects who ended up at Cal COE are MUCH weaker students. And I would go far and say that I’m inclined to believe that a lot of them are academically stronger than PiperXP, if she’s the benchmark of an MIT peep. </p>
PiperXP, I may not have four legs like a pig, but I can carry myself well in this debate. And I’m not ■■■■■■■■ - you know that. </p>
<p>I questioned the claim that Berkeley CEO peeps are MUCH weaker peeps of MIT. I think that was rather a false, condescending claim, and I explained why I thought so. How’s that categorised as ■■■■■■■■? </p>
<p>Perhaps, HYPSM+Caltech students have slightly higher SAT scores than Berkeley COE peeps, but not significantly higher, specially in the maths part of SATs. </p>
<p>I can’t back this up, but I think Berkeley COE peeps have comparable stats with Duke, Penn, Columbia and NU engineers. And, I’d go far and say, that, in the fields of engineering, Berkeley engineers are more accomplished and are more respected. </p>
<p>Aside to molliebatmit: I meant “good luck” with the new grant proposal. It’s too late to edit the post. We have a dog named “Lucky” so I suppose I just went along a well-worn neural groove.</p>
<p>People have pointed to objective data that you haven’t refuted, you claim to refute them, and you claim to have brought new objective data to light to prove your point when you have not.</p>
<p>As I said from the very beginning, “much” is subjective. Having a different opinion from yours is not the same as being condescending. </p>
<p>And now - until you actually bring up new data - I’ll ignore you :)</p>
<p>Thanks, QM. It’s pretty ugly out there in the sciences right now, and I feel like we’re throwing applications all over and not getting much in the way of money. At this rate, we’re going to publish a paper on our project without ever having been able to convince funding agencies that the thing is possible in principle.</p>
<p>As for in-state prestige of UCB vs. UMich/GT, I totally agree. But given the sheer number of California residents who apply to (and are admitted to) MIT, relative to other states, I am still surprised. I guess the numbers might even out a bit if you consider much of the Midwest to be “in-state” for UMich (not based on cost of attendance, obviously, but based on something more like regional identity) and much of the South to be “in-state” for GT. </p>
<p>I’m from Ohio, and certainly it was a badge of pride for students from my high school to study engineering at Michigan, even if it was in Michigan, our ancient and deadly enemy.</p>
To be fair, the word “significantly” doesn’t say anything about the magnitude of a difference, just that the difference is statistically likely to be real.</p>
I’m now applying (and applying, and applying) for grants and postdoc fellowships in my new lab.</p>